Artificial Intelligence in Supporting Early Intervention for Children with Autism Spectrum Disorder: Opportunities and Challenges

Abstract

Early identification and intervention are central to improving developmental trajectories for children with autism spectrum disorder (ASD), yet access to specialist assessment and therapy remains uneven. Artificial intelligence (AI) has been proposed as a way to extend the reach and precision of early support. This article reviews the opportunities and challenges of AI in early intervention for children with ASD, using a conceptual-analytical narrative review. It synthesizes peer-reviewed studies and authoritative guidance across three stages of the care pathway: screening, diagnostic decision support, and intervention delivery, and examines the limitations that accompany each. The findings indicate that machine learning applied to parent reports, home videos, eye-gaze and attention measures, and neuroimaging shows promising accuracy for early detection; that social robots and wearable aids can elicit social behavior in structured settings; and that these gains are constrained by small and unrepresentative samples, limited external validation, algorithmic bias, opacity, privacy risks, and uncertain transfer to everyday life. The discussion argues that AI is best positioned as an augmentation of, and not a substitute for, clinical judgment and family-centered care, and it outlines implications for practitioners and research priorities, including validation in culturally and linguistically diverse populations.